On Thu 17-12-15 16:15:21, Andrew Morton wrote:
> On Tue, 15 Dec 2015 19:36:15 +0100 Michal Hocko <[email protected]> wrote:
>
> > This patch reduces the probability of such a lockup by introducing a
> > specialized kernel thread (oom_reaper)
>
> CONFIG_MMU=n:
>
> slub.c:(.text+0x4184): undefined reference to `tlb_gather_mmu'
> slub.c:(.text+0x41bc): undefined reference to `unmap_page_range'
> slub.c:(.text+0x41d8): undefined reference to `tlb_finish_mmu'
>
> I did the below so I can get an mmotm out the door, but hopefully
> there's a cleaner way.
Sorry about that and thanks for your fixup! I am not very familiar with
!MMU world and haven't heard about issues with the OOM deadlocks yet. So
I guess making this MMU only makes some sense. I would just get rid of
ifdefs in oom_kill_process and provide an empty wake_oom_reaper for
!CONFIG_MMU. The following on top of yours:
---
diff --git a/mm/oom_kill.c b/mm/oom_kill.c
index 4b0a5d8b92e1..56ff1ff18c0e 100644
--- a/mm/oom_kill.c
+++ b/mm/oom_kill.c
@@ -537,6 +537,10 @@ static int __init oom_init(void)
return 0;
}
module_init(oom_init)
+#else
+static void wake_oom_reaper(struct mm_struct *mm)
+{
+}
#endif
/**
@@ -648,9 +652,7 @@ void oom_kill_process(struct oom_control *oc, struct
task_struct *p,
unsigned int victim_points = 0;
static DEFINE_RATELIMIT_STATE(oom_rs, DEFAULT_RATELIMIT_INTERVAL,
DEFAULT_RATELIMIT_BURST);
-#ifdef CONFIG_MMU
bool can_oom_reap = true;
-#endif
/*
* If the task is already exiting, don't alarm the sysadmin or kill
@@ -743,7 +745,6 @@ void oom_kill_process(struct oom_control *oc, struct
task_struct *p,
continue;
if (is_global_init(p))
continue;
-#ifdef CONFIG_MMU
if (unlikely(p->flags & PF_KTHREAD) ||
p->signal->oom_score_adj == OOM_SCORE_ADJ_MIN) {
/*
@@ -754,15 +755,12 @@ void oom_kill_process(struct oom_control *oc, struct
task_struct *p,
can_oom_reap = false;
continue;
}
-#endif
do_send_sig_info(SIGKILL, SEND_SIG_FORCED, p, true);
}
rcu_read_unlock();
-#ifdef CONFIG_MMU
if (can_oom_reap)
wake_oom_reaper(mm);
-#endif
mmdrop(mm);
put_task_struct(victim);
--
Michal Hocko
SUSE Labs
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
